Diastereotopic Protons and Complex Splitting Patterns

Diastereotopicity is not a museum-piece concept; it is load-bearing in modern structure elucidation and drug development. In total synthesis and natural-product structure determination, a diastereotopic methylene's pattern of couplings is often the only direct evidence for the relative configuration of a nearby stereocenter, especially when no NOE contact is available; medicinal chemists routinely use the appearance (or collapse) of diastereotopicity in a CH₂ adjacent to a newly installed stereocenter as a quick, non-destructive check that asymmetric synthesis actually worked, since a racemic or epimeric mixture gives subtly different (or doubled) diastereotopic patterns compared to a single diastereomer. In protein NMR, the $\beta$-methylene protons of nearly every amino acid side chain (except glycine and achiral side chains) are diastereotopic because of the adjacent $\alpha$-stereocenter, and their distinct chemical shifts and $J_{\alpha\beta}$ couplings are used routinely in structure calculations to constrain side-chain rotamer populations via exactly the Karplus-equation logic developed above. In polymer and materials NMR, diastereotopic methylene splitting reports on tacticity — whether neighboring stereocenters along a polymer backbone are isotactic, syndiotactic, or atactic — because the local environment of a backbone $\mathrm{CH_2}$ depends on the configuration of its neighbors up and down the chain. And whenever a diastereotopic pattern becomes too congested or too strongly-coupled to disentangle by hand splitting-trees — precisely the second-order warning raised in Mechanism 2 — the modern solution is two-dimensional correlation spectroscopy, which spreads overlapping couplings into a second frequency dimension and reads off which proton couples to which directly from cross-peaks, the natural next tool once the one-dimensional bookkeeping developed here reaches its limits, as introduced in intro-2d-nmr-cosy.

Ask the tutor: How would a COSY spectrum let me assign Jax and Jbx directly, without building the splitting tree by hand?

Begin with the substitution test itself, applied mechanistically rather than declared by fiat. Take a carbon $\mathrm{CH_2XY}$ bonded to two hydrogens $H_a$ and $H_b$. Imagine replacing $H_a$ by a test nucleus — deuterium, or an isotopic label, it does not matter — leaving $H_b$ untouched; call the product molecule P_a. Now instead replace $H_b$, leaving $H_a$ untouched; call this product P_b. Compare P_a and P_b:

• If P_a and P_b are the same compound (superimposable, identical connectivity and identical stereochemistry), $H_a$ and $H_b$ are homotopic. This happens exactly when a proper rotation axis $C_n$ of the molecule swaps them.
• If P_a and P_b are non-superimposable mirror images of each other (enantiomers), $H_a$ and $H_b$ are enantiotopic. This happens when only an improper symmetry element (a mirror plane $\sigma$, an $S_n$ axis, or a center of inversion) relates them.
• If P_a and P_b are diastereomers — different constitutions are impossible here since both are $\mathrm{CHDXY}$, but they can differ in the relative configuration between the new stereocenter at the former $\mathrm{CH_2}$ and any other stereocenter in the molecule — then $H_a$ and $H_b$ are diastereotopic.

The crucial mechanistic insight is why diastereomers have different NMR properties while enantiomers do not. A pair of enantiomers has, atom for atom, identical scalar and distance relationships to any achiral probe: the same bond lengths, same bond angles, same through-space distances to every other nucleus, because reflection preserves all of these scalar quantities. An achiral NMR experiment (ordinary solvent, ordinary spectrometer) cannot distinguish mirror images, so enantiotopic protons resonate at literally the same frequency and couple identically to everything else — one signal. A pair of diastereomers, by contrast, has different internal geometries: different dihedral angles, different through-space contacts, different populations of rotamers, because they are not mirror-related at all — they are different molecules that happen to share a molecular formula and most of their connectivity. Different geometry means different shielding (different $\delta$) and, via the Karplus relationship between dihedral angle and three-bond coupling, different $J$ values to a common neighboring proton. This is the mechanistic bridge to splitting patterns. Consider a fragment $H_a\text{–}C\text{–}C^*(H_x)$ where $C^*$ bears a stereocenter (or, more generally, where the local conformational average is not symmetric between $H_a$ and $H_b$). Rotation about the $C\text{–}C$ bond populates staggered rotamers, and in each rotamer $H_a$ sits at some dihedral angle $\phi_{aX}$ to $H_X$ while $H_b$ sits at a different dihedral angle $\phi_{bX}$ — they cannot be related by rotation because $H_a$ and $H_b$ are attached to a prochiral carbon flanked by an asymmetric neighbor. The Boltzmann-weighted average of $\cos^2\phi$ over the accessible rotamers is therefore different for the $a$–X and $b$–X pathways, and by the Karplus equation this forces

even though both couplings are nominally "three-bond, freely rotating." This is the mechanistic seed of every ABX pattern in this article: not a mysterious splitting rule, but ordinary vicinal coupling made asymmetric by molecular chirality.

The consequence for spectral appearance is that a diastereotopic $\mathrm{CH_2}$ typically shows two chemical shifts $\delta_a \neq \delta_b$, a large geminal coupling $J_{ab} = {}^2J_{HH}$ (typically $-10$ to $-15\,\text{Hz}$ for sp³ CH₂ flanked by no π system, more negative approaching $-12$ Hz for simple alkyl chains), and, if a vicinal proton $H_X$ is present, two different vicinal couplings $J_{aX} \neq J_{bX}$. The spin system is no longer "$AX_2$" (one shift for two equivalent protons) but genuinely ABX (three distinct shifts, three distinct coupling constants $J_{ab}, J_{aX}, J_{bX}$), and each of $H_a$, $H_b$ appears as a doublet of doublets rather than the naive doublet the $n+1$ rule would predict for "CH₂ next to CH."

The governing relationships are three, and together they replace "count neighbors, apply $n+1$" with a rigorous, structure-sensitive bookkeeping scheme.

1. The symmetry criterion for equivalence. Two nuclei $H_a, H_b$ are chemical-shift equivalent if and only if there exists a symmetry operation of the molecule (a proper rotation for homotopic, or the achiral-medium-invisible improper operation for enantiotopic) that maps $H_a \to H_b$:

where: $\hat O$ ranges over the molecule's symmetry operations; a proper rotation gives true magnetic identity in any medium, while an improper operation gives identity only when the observing medium is itself achiral.

2. The Karplus dependence that makes diastereotopic vicinal couplings differ. For a $H\text{–}C\text{–}C\text{–}H$ fragment,

where: $\phi$ is the H–C–C–H dihedral angle, and $A \approx 8$–$10\,\text{Hz}$, $B\approx -1\,\text{Hz}$, $C\approx 0$–$1.5\,\text{Hz}$ are empirical constants (values shift with electronegative substituents and ring strain). Because $H_a$ and $H_b$ on a prochiral $\mathrm{CH_2}$ occupy different average dihedral angles to a vicinal proton $H_X$ when a nearby stereocenter breaks the local mirror symmetry, the population-weighted couplings differ:

where: $p_i$ is the Boltzmann population of rotamer $i$, and $\phi_{aX,i}, \phi_{bX,i}$ are the dihedral angles in that rotamer — necessarily unequal because the stereocenter removes the local $C_s$ symmetry that would otherwise force $\phi_{aX}=\phi_{bX}$ (mod sign).

3. Line count and pattern for a first-order ABX (or ABMX) system, replacing $n+1$:

where: $k$ is the number of distinct nonzero coupling partners resonating at a different shift from the observed proton (for a doublet-of-doublets, $k=2$; for a ddd, $k=3$), and lines may overlap ("collapse") if two $J$ values are numerically equal or very close, or may merge into a strongly-coupled second-order pattern if $\Delta\delta$ (in Hz) is not large compared to $J$ — the second-order regime discussed further below.

Reading the three relationships together

The first equation is the gatekeeper: it tells you, before you even draw a structure, whether two protons can possibly be equivalent. The second equation is the mechanistic engine: it explains why, once two protons are diastereotopic, their couplings to a third nucleus are numerically different rather than accidentally similar. The third equation is the bookkeeping consequence: with $k$ inequivalent coupling partners you get up to $2^k$ lines per proton, not the compressed $n+1$ count that assumes all couplings to a given proton are identical.

Example 1 — Foundational: styrene oxide, the textbook ABX epoxide

Styrene oxide, $\mathrm{C_6H_5\text{–}CH(O)CH_2}$ (a three-membered epoxide ring with a phenyl on the carbon bearing $H_X$), is the standard teaching example precisely because the epoxide ring locks the geometry — there is no free rotation to average anything, so the $J$ values are large and unambiguous.

Assignment of the spin system. The ring carbon bearing phenyl carries $H_X$ (the "methine," $\delta_X \approx 3.85$ ppm); the terminal ring $\mathrm{CH_2}$ carries $H_a$ (trans to phenyl across the ring, $\delta_a \approx 3.14$ ppm) and $H_b$ (cis to phenyl, $\delta_b \approx 2.80$ ppm). The stereocenter at the phenyl-bearing carbon makes $H_a$ and $H_b$ diastereotopic by the substitution test: replacing $H_a$ by D gives the $(2R,\;\text{new-}S)$ epoxide-d₁ while replacing $H_b$ gives the $(2R,\;\text{new-}R)$ epoxide-d₁ (using the ring stereocenter as reference) — these are diastereomers, confirming diastereotopicity directly, independent of any conformational argument (the ring geometry is rigid, so this is the cleanest possible demonstration).

The coupling constants (all values well-established from decades of epoxide NMR studies).

Note immediately: the geminal coupling in the epoxide ($-5.6$ Hz) is far smaller in magnitude than a typical open-chain geminal coupling ($-12$ to $-15$ Hz) — this itself is a mechanistic consequence of ring strain changing the hybridization at the $\mathrm{CH_2}$ carbon (more p-character in the C–H bonds of a strained three-membered ring lowers the magnitude of ${}^2J_{HH}$, a well-documented ring-size effect on geminal coupling, running from roughly $+5$ to $+7$ Hz in cyclopropanes/epoxides up to $-12$ to $-15$ Hz in ordinary acyclic and six-membered-ring methylenes, with the sign changing along the way).

Building H_a's pattern. $H_a$ is split by $J_{ab}=5.6$ Hz and by $J_{aX}=4.0$ Hz, giving a doublet of doublets:

four lines, spacing pattern $(0.8, 3.2, 0.8)$ Hz reading outward-in — a visibly asymmetric dd, nothing like a clean triplet.

Building H_b's pattern. $H_b$ split by the same $J_{ab}=5.6$ Hz and by $J_{bX}=2.6$ Hz:

Limiting-case check. If, hypothetically, the phenyl group were replaced by a second hydrogen (turning the molecule into plain ethylene oxide), the ring carbon bearing $H_X$ would no longer be a stereocenter — the molecule would gain a $C_{2v}$ symmetry with both a $C_2$ axis and mirror planes relating $H_a$ and $H_b$. The substitution test would then give homotopic (or at least equivalent) protons, $J_{aX}$ would necessarily equal $J_{bX}$ by symmetry, and the pattern would collapse to a single symmetric signal — exactly matching the well-known fact that ethylene oxide shows one sharp ¹H singlet-like resonance. This confirms the formula: removing the asymmetric substituent removes diastereotopicity, and the "different $J$" phenomenon vanishes exactly as the symmetry argument requires.

Example 2 — Applied: the exocyclic CH₂ of citronellal (an acyclic stereocenter case)

Citronellal, $(R)$-3,7-dimethyloct-6-enal, contains a stereocenter at C3 flanked directly by the aldehyde-bearing $\mathrm{CH_2}$ at C2. This is the "textbook acyclic" diastereotopic case, and the one most often first confusing students because there is no ring to make the geometry obviously rigid — the diastereotopicity survives despite free rotation, exactly per Mechanism 1.

Spin system. $\mathrm{OHC(H_X)\text{–}CH_aH_b\text{–}C_3^*H(CH_3)\text{–}}$(chain). Here $H_X$ is the aldehyde proton, resonating far downfield ($\delta \approx 9.75$ ppm), so it is cleanly resolved from the $\mathrm{CH_2}$ at $\delta \approx 2.2$–$2.4$ ppm — a large $\Delta\delta$, safely first-order.

Typical literature-consistent values. $\delta_a = 2.38$ ppm, $\delta_b = 2.22$ ppm, $J_{ab} = -16.0\,\text{Hz}$ (large geminal, typical of an acyclic CH₂ adjacent to a carbonyl, where the $\pi$ system increases the magnitude of ${}^2J$ relative to a plain alkyl $\mathrm{CH_2}$), $J_{aX} = 2.8\,\text{Hz}$, $J_{bX} = 2.2\,\text{Hz}$ (small, because free rotation about the $\mathrm{C2\text{-}C3}$ bond averages the aldehyde coupling toward the small end of the Karplus curve; note the two vicinal couplings are close but not equal — a hallmark of a remote, less sterically decisive stereocenter, illustrating that diastereotopicity does not guarantee a large $\Delta J$, only a nonzero one).

Mechanistic reading. By the logic of Mechanism 1, the three staggered rotamers about $\mathrm{C2\text{-}C3}$ are non-degenerate because C3 bears $\mathrm{CH_3}$, H, and the isoprenoid chain — three different groups — so $J_{aX}\ne J_{bX}$ generically. The near-closeness of the two values ($2.8$ vs $2.2$ Hz) reflects that the energy differences among the three rotamers here are modest (no strong 1,3-interactions or polar/steric bias), so the population-weighted averages happen to land close together — but "close" is not "equal," and the two aldehyde-proton couplings remain resolvably different in a well-resolved 400+ MHz spectrum, producing a very slightly asymmetric doublet-like fine structure on the aldehyde triplet-of-triplets-looking signal that a careless reader might mistake for a clean triplet.

Limiting case. If citronellal were replaced by an achiral analogue lacking the C3 methyl (i.e., an unbranched aldehyde $\mathrm{OHC\text{-}CH_2\text{-}CH_2\text{-}R}$ with no stereocenter at C3), the local mirror symmetry through the $\mathrm{CH_2}$ would be restored, forcing $J_{aX}=J_{bX}$ exactly and collapsing the aldehyde signal to a clean triplet — precisely the behavior seen for unbranched aliphatic aldehydes, confirming that the C3 stereocenter, not the aldehyde itself, is the diastereotopicity-inducing feature.

Example 3 — Exam/competition level: a 4-substituted cyclohexanone axial/equatorial CH₂, second-order complication included

Consider 4-tert-butylcyclohexan-1-one substituted at C2 with a single deuterium so as to leave one stereocenter-adjacent $\mathrm{CH_2}$ intact at C3 — or, more realistically for exam purposes, consider the C2/C6 methylenes flanking the carbonyl in any 4-substituted cyclohexanone bearing a stereocenter elsewhere in the ring (e.g., 4-methylcyclohexanone is achiral and not useful here; instead take (4R)-4-phenylcyclohexan-1-one is still achiral overall by symmetry — so the pedagogically sharp case is a 3-substituted cyclohexanone, e.g. 3-methylcyclohexan-1-one).

Why this is the hard case. The bulky tert-butyl (or here, the C3 methyl) locks the ring into a single dominant chair, so the two protons at C2 (flanking both the carbonyl and the stereocenter at C3) are now diastereotopic and held in fixed axial/equatorial positions — not freely rotating, ring-flip-averaged positions. This is the scenario where diastereotopic splitting is at its most dramatic, because axial and equatorial protons on a locked chair have very different vicinal dihedral angles to their ring neighbors (axial–axial $\approx 180°$, giving large $J\approx 10$–$13$ Hz; axial–equatorial and equatorial–equatorial $\approx 60°$, giving small $J\approx 3$–$5$ Hz) and different geminal environments, all superimposed on genuine diastereotopicity from the C3 stereocenter.

Spin system at C2 (using generic ring-locked values). Label the two C2 protons $H_{2ax}$ and $H_{2eq}$ (diastereotopic — the substitution test again gives diastereomers, since C3 is a stereocenter), and the two C3-adjacent-neighbor protons at C4... but to keep this tractable at exam level, focus on the well-defined three-spin fragment $H_{2ax}$–$H_{2eq}$–$H_{3}$ (C3 bearing the methyl and the ring stereocenter, with one ring H at C3):

Step-by-step line construction for $H_{2ax}$ (analogous to Mechanism 2, Step 2): split first by $J_{2ax,2eq}=13.0$ Hz into two lines at $\pm 6.5$ Hz, each further split by $J_{2ax,3}=11.5$ Hz into $\pm 5.75$ Hz sub-splittings:

— a widely-spaced dd, easily recognized as "axial" by its large outer splitting (this large-J-large-J pattern, historically, is exactly how chemists assign axial vs equatorial protons in locked cyclohexanones by inspection).

For $H_{2eq}$: split by the same $J_{2ax,2eq}=13.0$ Hz (±6.5 Hz) then by the small $J_{2eq,3}=4.2$ Hz (±2.1 Hz):

— a visibly "tighter" dd, the classic equatorial-proton fine-structure signature (large geminal splitting but small vicinal fine splitting, because equatorial–axial dihedral angles are near $60°$).

Where n+1 fails completely here, and the second-order warning. A student applying $n+1$ blindly to "$\mathrm{CH_2}$ next to one $\mathrm{CH}$" would predict a single doublet (n = 1 neighbor) for "the $\mathrm{CH_2}$ signal" — utterly wrong on three counts: (i) there are two chemical shifts, not one; (ii) each shows a doublet of doublets, not a doublet, because the geminal partner (at a different, resolved shift) also splits it; (iii) the two vicinal couplings to $H_3$ differ by nearly a factor of three ($11.5$ vs $4.2$ Hz), immediately diagnostic of the axial/equatorial relationship — information the $n+1$ rule is structurally incapable of returning, since it assumes indistinguishable neighbors. If, in a less rigid ring or at elevated temperature, ring-flipping became fast on the NMR timescale, $H_{2ax}$ and $H_{2eq}$ would exchange environments (each spending equal time axial and equatorial), and their observed couplings would collapse toward a common motionally averaged value $\bar J = \tfrac{1}{2}(J_{2ax,3}+J_{2eq,3}) \approx 7.85$ Hz for both — this is the dynamic-averaging limiting case, and it is distinct from (and must not be confused with) true symmetry-enforced equivalence: the two protons remain constitutionally/configurationally diastereotopic even when fast exchange makes them appear equivalent by time-averaging, exactly as two enantiomeric conformers interconverting fast can average an NMR signal without the underlying stereochemical distinction disappearing.

The conceptual vocabulary of prochirality and topicity — homotopic, enantiotopic, diastereotopic — was formalized in the mid-twentieth century primarily through the work of Kurt Mislow, who systematized the substitution-test logic used throughout this article, building on earlier stereochemical language from Kasimir Fajans and the broader tradition of dynamic stereochemistry established by van't Hoff and Le Bel for static configurations, then extended by Mislow and coworkers to prochiral relationships specifically relevant to spectroscopy. The application to real NMR splitting patterns — recognizing that diastereotopic methylenes routinely give ABX and ABMX patterns rather than clean multiplets — became common laboratory knowledge through the 1960s and 1970s as high-resolution NMR spectrometers made small $\Delta\delta$ and $\Delta J$ differences experimentally visible for the first time, turning what had been a theoretical nicety into an everyday diagnostic tool.